Not able Convert -22V to adjustable -3V to -16V 10A using lm337 and tip2955

[khuram9172]

I have simulated a circuit on the proteus to convert -22V into adjustable -3V to -16V using LM337 and Tip2955. In the simulation when I change the load, output voltage across it remain stable but when I built the hardware I am having a voltage drop of 1 to 2 volts from the adjusted voltage and also tip2955 is heating like crazy at 5A. Can someone please tell me what mistake I am making or if there is any other way to achieve this or what adjustments I can make in this circuit to achieve the desired results. In simulation it looks fine but in reality it doesn't work properly. Picture of my circuit is attached here:

 

[betwixt]

You are using the LM337 as a bias source then the transistors as a current amplifier. I would be better to add a resistor in series with the INPUT of the LM337 and use that as the bias source, that means the load current is shared between all devices and you get the benefit of the overheat and overcurrent protection as well.
The transistors would have to be replaced with NPN types. I would advise placing a small ( a few ohms) resistor in the base connection of each transistor to balance the current through them, at the moment the one with lowest Vbe hogs the current from the others. Also you should have capacitors at the input pin and preferable also the output pin of the regulator.

Brian.

 

[KlausST]

Hi,

where is this schematic from?
* from any internet source (web site, datasheet, video...) --> please post a link
* your own?

and also tip2955 is heating like crazy at 5A.

You can´t change physics:
For any linear dropping the dissipated power (heat) is identical: It´s (V_in - V_ou) * I. No matter if you use a BJT, MOSFET, diode, resistor ....
(the dissipation of the quiescent current (GND current) can be ignored here)

For sure if you drop from 22V to 3V .. you "waste" 18V. 18V * 5A = 90W of heat. That´s a lot.

If you don´t want this heat --> change your design to a switch mode design. .. it would probably dissipate less than 10W of heat .. more advanced designs maybe less than 3W.

Klaus

 

[khuram9172]

You are using the LM337 as a bias source then the transistors as a current amplifier. I would be better to add a resistor in series with the INPUT of the LM337 and use that as the bias source, that means the load current is shared between all devices and you get the benefit of the overheat and overcurrent protection as well.
The transistors would have to be replaced with NPN types. I would advise placing a small ( a few ohms) resistor in the base connection of each transistor to balance the current through them, at the moment the one with lowest Vbe hogs the current from the others. Also you should have capacitors at the input pin and preferable also the output pin of the regulator.

Brian.

So, you are saying if I make a circuit like this as I have shown below, the transistors will not generate heat and the output voltage will be stable on changing load.

 

[khuram9172]

Hi,

where is this schematic from?
* from any internet source (web site, datasheet, video...) --> please post a link
* your own?


You can´t change physics:
For any linear dropping the dissipated power (heat) is identical: It´s (V_in - V_ou) * I. No matter if you use a BJT, MOSFET, diode, resistor ....
(the dissipation of the quiescent current (GND current) can be ignored here)

For sure if you drop from 22V to 3V .. you "waste" 18V. 18V * 5A = 90W of heat. That´s a lot.

If you don´t want this heat --> change your design to a switch mode design. .. it would probably dissipate less than 10W of heat .. more advanced designs maybe less than 3W.

Klaus

This is my own schematics, designed this on proteous. Well may be you are right about designing a switch-mode power supply. As I haven't designed a negative switch mode power supply before or even a positive one (still learning), so can you give a link or a circuit design which can fulfill my requirements. Thank you!

 

[FvM]

So, you are saying if I make a circuit like this as I have shown below, the transistors will not generate heat and the output voltage will be stable on changing load.

Using npn transistors is necessary to achieve correct voltage regulation. The circuit is o.k. except for 0.2 ohm base resistors that are useless in this position. For current balancing, you would place emitter series resistors, 0.05 to 0.2 ohm. Heat dissipation is however intrinsic to a linear regulator and can't be avoided, as explained by KlausST. Total ploss = (Vin - Vout)*Iout. You need a large heatsink to dissipate about 19V* 10A = 190 W. It's also essential to place LM337 on the same heatsink to protect the transistors against overheating.

--- Updated Today at 10:22 AM ---

It should be noted that the circuit has no short circuit protection. A sudden output short can destroy the power transistors by exceeding safe operation area.

--- Updated Today at 10:26 AM ---

Decision between linear and switch-mode regulator is a trade-off, depending on application, costs, room etc.

 

[KlausST]

As I haven't designed a negative switch mode power supply before or even a positive one (still learning), so can you give a link or a circuit design which can fulfill my requirements. Thank you!

* you can try to buy a module (I did not do a search)

* you can build one on your own. There are semiconductor manufacturers that have switch mode ICs for your application. They provide all you need. like: a selection guide, datasheets, application notes, design notes, online schematic design, simulators ... maybe even videos.

As with all electronic parts: careful reading, understanding and following their informations is essential.

Klaus

 

[danadakk]

You have the following considerations (and others in datasheet) for the LM337 :

Rough equation for efficiency :

This reduces ~ to (Vout/Vin) x 100%. so very inefficient at low output voltages. Not exactly
a green approach to the problem.

Switch mode, as previously recommended, can achieve > 80%, typical 85% to low 90%. So
if you want to build a room heater stay with linear approach.


Regards, Dana.